1. Field of the Invention
The present invention relates to an apparatus and a method for electron beam lithography having a beam-size-correcting function and a semiconductor device fabricated using the method for writing.
2. Description of the Related Art
Techniques of electron beam lithography are attracting attention as techniques for forming desired fine patterns on the surface of a sample such as semiconductor wafers. In the case of the formation of a large scale circuit pattern for which high throughput is needed, there is a trend toward methods for drawing a circuit pattern by dividing it into rectangles and trapezoids in different sizes (variable shaping lithography) and methods for drawing a plurality of such graphic patterns together (simultaneous shaping lithography).
There are conventional apparatuses for electron beam lithography which comprise a first aperture mask formed with a shaping window and a second aperture mask formed with a variable shape window and a simultaneous shaping lithography to perform both the variable shaping lithography and the simultaneous shaping lithography.
When the variable shaping lithography is used in such an apparatus for electron beam lithography, electron beams are applied to the entire surface of the shaping window of the first aperture mask, and an electron beam image from the shaping window is projected upon the variable shaping window of the second aperture mask in a displaced position to form an electron beam image in a different size which is used to draw a pattern of arbitrary size. When the simultaneous shaping lithography is used, electron beams are applied to the entire surface of the shaping window of the first aperture mask, and an electron beam from the shaping window is projected upon the simultaneous shaping window of the second aperture mask in a displaced position to form an electron beam image consisting of a plurality of shapes which is used to draw a desired pattern.
In such an apparatus for electron beam lithography using the variable shaping or simultaneous shaping lithography, an actual beam size can include an error from a set value caused by a displacement of the axis of the optical system for electron beams, a rotational bias of a deflector, an electrical effect of the circuit and the like, which results in a need for a technique for measuring and correcting a beam size with high accuracy.
In such apparatuses for electron beam lithography, a beam size has been measured using techniques such as a knife-edge method and has been corrected based on the result of such measurement. The knife-edge method is a technique wherein a detection mark is made of a metal such as gold having a high reflection factor in a predetermined position in the vicinity of a sample; an electron beam is scanned across an edge of the detection mark; an electron signal reflected from the detection mark is differentiated; and 50% of the strength of the resultant curve is determined as the beam size of the electron beam.
The knife-edge method described above provides a stable result in regions having relatively large dimensions such as 1.0 .mu.m or more. However, in fine regions in which the beam size is less than 1.0 .mu.m, using the knife-edge method, reflected electrons and current amount are smaller than those in relatively large regions of 1.0 .mu.m or more, which results in a reduced S/N ratio and consequently results in reduced accuracy in correcting a beam size. For this reason, a beam size in such a fine region has been obtained by performing extrapolation (linear approximation) based on the result of measurement in an interpolated region (of dimensions in the range from 1.0 to 5.0 .mu.m) wherein the beam size can be measured with stability and by making correction based on the result of the extrapolation. In fine regions where the beam size is less than 1.0 .mu.m, however, astigmatic components (blur) occupy a relatively large part of a beam size. This results in a problem in that the use of the method based on linear approximation does not provide high accuracy because a correction error still remains.
A current density method is a method for correcting a beam size obtained by the knife-edge method. The current density method is a method wherein current amount is measured while varying each of set values of longitudinal and lateral width of a beam size in a dimensional range less than 1.0 .mu.m on the basis of a beam size in the range from 1.0 to 5.0 .mu.m obtained by the knife-edge method as the reference and wherein a correction table is created for each size under a condition that the beam size and current amount are proportional to each other (i.e., the current density is constant). The current amount is measured by fixing either the longitudinal or lateral width of rectangular beam patterns and by varying the other stepwise, and the one-dimensional relationship between measured values obtained from the current amount and set values is identified. Correction is made after the measurement using polygonal line correction wherein a beam size is corrected with a polygonal line such that a constant current density is achieved or wherein a beam size is shifted in certain amounts such that a constant current density is achieved.
In the case of the method to create a table of correction values in which measured values and set values are in one-dimensional relationship with each other, the measurement of a beam size and current amount in fine regions smaller than 1.0 .mu.m is vulnerable to quantization errors in digital circuits and power supply noises on analog circuits which are directly reflected on the correction table. Therefore, the conventional method has a problem in that it cannot achieve accuracy required for a beam size correction in regions finer than 1.0 .mu.m. In addition, conventional apparatuses for electron beam lithography have had dimensional differences between patterns caused by a phenomenon wherein actually formed beams have different lateral widths, even if the lateral width of set value is constant when the beams have different longitudinal widths (hereinafter referred to as "longitudinal dependence of a beam size" .) The conventional method for correcting a beam size described above has a problem in that it can not eliminate such longitudinal dependence of a beam size.